A couple of points about the use of performance to identify "flags":
First, I'd really love to see the specifics of the studies that show how 9W/kg is possible, Andy? Apart from referring to this literature, what is the physiology behind that? The Joyner paper on marathon running provided only the explanation of those factors which are already used by the likes of Vayer and myself to estimate the physiological demand of certain cycling performances - cycling efficiency (analogous to Joyner's economy), VO2max and lactate threshold (which is a driver of the intensity that a rider can sustain for a given period). These are exactly what are used to make the calculations, so the Joyner paper does little to show the flaws (unless you're criticizing the Joyner paper...?)
So, having already taken those into account, people like Vayer, me and Pierre Salet (all quoted in the New Scientist article) are estimating a limit to performance. The Joyner paper does nothing to highlight "limitations", and again, if there is a paper I am missing that shows that 9W/kg is possible, then please provide it rather than just alluding to it. I realize I may still be a "newly-minted" physiologist (to use your description of me) but I'd like to see the facts, not assertions.
With regards to the error, you're 100% correct. There is error. That is why nobody has yet said that this kind of analysis "PROVES" doping, only that it can highlight cases that may be suspected. Anyone who says "I've proved doping based on performance" has a problem, and deserves criticism. But to say you're embarrassed for the field, that seems a bit extreme because a) the author of that piece is not from your field so surely one should be a little more constructive - write to him and show him the "light", and b) it's an over-reaction to an idea - the whole tone of the New Scientist piece is very speculative, as it should be.
And just a note on the error - in a study done by Portoleau et al, the estimation method was validated against SRM measurements in 16 male cyclists, and the average difference between the two was 0.24% (CI = -6.1, 6, 6%).
Where the larger error comes in, and I am stating this openly because I don't wish to purport this level of accuracy, is when you estimate rider mass, and also some of the assumptions you make regarding the cyclist's efficiency - is efficiency 23% or 25%? I go with 23% myself, because that's what the best Tour rider in recent years was measured at, and because that's what we get in our lab on Tour-level riders. Note also that as a cyclist's efficiency rises, their VO2max comes down (Lucia et al., http://www.ncbi.nlm.nih.gov/pubmed/12471319
). So a rider with efficiency of 21% is much more likely to have a VO2max of 86ml/kg/min, whereas someone at 23% will achieve the same performance with a VO2max of say 78ml/kg/min. The combination of high efficiency AND high maximal O2 consumption is very rare indeed.
But in my view, the error in the data is much smaller than the differences in physiology that would be expected. I see Prof Hans Rosling's population data as an example of this - they are collecting demographic data from all over the world and are roundly criticized for the lack of "quality control" on the data, but his point, which I agree with, is that the differences in the data are larger than the error in the data, so they are still enormously valuable.
Now, I would love to see the physiology of 9W/kg. But outright dismissals of ideas, that's not science, and I don't think it helps people at all. So if you know how 9W/kg can be achieved:
a) Get yourself to France right now, because there are millions to be made!, and
b) Inform us all!